| Guido van Rossum | 9c30c24 | 1998-12-22 05:19:29 +0000 | [diff] [blame] | 1 | #!/usr/bin/python |
| 2 | # Module and documentation by Eric S. Raymond, 21 Dec 1998 |
| 3 | |
| 4 | import sys, os, string |
| 5 | |
| 6 | class shlex: |
| 7 | "A lexical analyzer class for simple shell-like syntaxes." |
| 8 | def __init__(self, instream=None): |
| 9 | if instream: |
| 10 | self.instream = instream |
| 11 | else: |
| 12 | self.instream = sys.stdin |
| 13 | self.commenters = '#' |
| 14 | self.wordchars = 'abcdfeghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_' |
| 15 | self.whitespace = ' \t\r\n' |
| 16 | self.quotes = '\'"' |
| 17 | self.state = ' ' |
| 18 | self.pushback = []; |
| 19 | self.lineno = 1 |
| 20 | self.debug = 0 |
| 21 | self.token = '' |
| 22 | |
| 23 | def push_token(self, tok): |
| 24 | "Push a token onto the stack popped by the get_token method" |
| 25 | if (self.debug >= 1): |
| 26 | print "Pushing " + tok |
| 27 | self.pushback = [tok] + self.pushback; |
| 28 | |
| 29 | def get_token(self): |
| 30 | "Get a token from the input stream (or from stack if it's monempty)" |
| 31 | if self.pushback: |
| 32 | tok = self.pushback[0] |
| 33 | self.pushback = self.pushback[1:] |
| 34 | if (self.debug >= 1): |
| 35 | print "Popping " + tok |
| 36 | return tok |
| 37 | tok = '' |
| 38 | while 1: |
| 39 | nextchar = self.instream.read(1); |
| 40 | if nextchar == '\n': |
| 41 | self.lineno = self.lineno + 1 |
| 42 | if self.debug >= 3: |
| 43 | print "In state " + repr(self.state) + " I see character: " + repr(nextchar) |
| 44 | if self.state == None: |
| 45 | return '' |
| 46 | elif self.state == ' ': |
| 47 | if not nextchar: |
| 48 | self.state = None; # end of file |
| 49 | break |
| 50 | elif nextchar in self.whitespace: |
| 51 | if self.debug >= 2: |
| 52 | print "I see whitespace in whitespace state" |
| 53 | if self.token: |
| 54 | break # emit current token |
| 55 | else: |
| 56 | continue |
| 57 | elif nextchar in self.commenters: |
| 58 | self.instream.readline() |
| 59 | self.lineno = self.lineno + 1 |
| 60 | elif nextchar in self.wordchars: |
| 61 | self.token = nextchar |
| 62 | self.state = 'a' |
| 63 | elif nextchar in self.quotes: |
| 64 | self.token = nextchar |
| 65 | self.state = nextchar |
| 66 | else: |
| 67 | self.token = nextchar |
| 68 | if self.token: |
| 69 | break # emit current token |
| 70 | else: |
| 71 | continue |
| 72 | elif self.state in self.quotes: |
| 73 | self.token = self.token + nextchar |
| 74 | if nextchar == self.state: |
| 75 | self.state = ' ' |
| 76 | break |
| 77 | elif self.state == 'a': |
| 78 | if not nextchar: |
| 79 | self.state = None; # end of file |
| 80 | break |
| 81 | elif nextchar in self.whitespace: |
| 82 | if self.debug >= 2: |
| 83 | print "I see whitespace in word state" |
| 84 | self.state = ' ' |
| 85 | if self.token: |
| 86 | break # emit current token |
| 87 | else: |
| 88 | continue |
| 89 | elif nextchar in self.commenters: |
| 90 | self.instream.readline() |
| 91 | self.lineno = self.lineno + 1 |
| 92 | elif nextchar in self.wordchars or nextchar in self.quotes: |
| 93 | self.token = self.token + nextchar |
| 94 | else: |
| 95 | self.pushback = [nextchar] + self.pushback |
| 96 | if self.debug >= 2: |
| 97 | print "I see punctuation in word state" |
| 98 | state = ' ' |
| 99 | if self.token: |
| 100 | break # emit current token |
| 101 | else: |
| 102 | continue |
| 103 | |
| 104 | result = self.token |
| 105 | self.token = '' |
| 106 | if self.debug >= 1: |
| 107 | print "Token: " + result |
| 108 | return result |
| 109 | |
| 110 | if __name__ == '__main__': |
| 111 | |
| 112 | lexer = shlex() |
| 113 | while 1: |
| 114 | tt = lexer.get_token() |
| 115 | if tt != None: |
| 116 | print "Token: " + repr(tt) |
| 117 | else: |
| 118 | break |
| 119 | |